Research On Visual Detection Technology Of Automotive Stamping Parts Scratches Based On Structured Light

Stamping parts make up a large proportion of the automobile manufacturing industry.As a common defect of automobile stamping parts,scratches will not only directly affect the overall safety and quality of the automobile but also lead to oxidation,corrosion,and other phenomena of the body parts,thus reducing the service life of the automobile.In this study,automobile stamping parts with scratches are taken as the target to be measured,a 3D point cloud acquisition platform based on a structured light sensor is built,and scratch extraction and measurement methods on the surface of automobile stamping parts are studied.Specific research contents are as follows:(1)Construction of a 3D point cloud acquisition platform.The structured light method based on laser triangulation is studied to obtain the surface topography information of automobile stamping parts with scratches.A 3D point cloud acquisition platform was built using the structured light sensor.The 3D point cloud data collected on the surface of automobile stamping parts was the core data of the scratch detection experiment.(2)Scratch point cloud localization.Based on the point cloud’s normal vector and curvature information,a scratch localization algorithm called bilateral weight integration is proposed,which can calculate the probability that each point in the point cloud belongs to the scratch.Through this probability information,we can determine the approximate location of the scratch to complete the scratch localization process.(3)Scratch point cloud complement.The NRPCA algorithm is proposed by combining the PCA(Principal Components Analysis)algorithm with the point cloud’s normal vector to achieve the scratches’ width and depth direction finding.Then,a scratch complement algorithm named NRPCA-RG is proposed by combining NRPCA and region growth algorithm,which can determine all the 3D points corresponding to the scratches with high efficiency and accuracy.After the scratch localization and complement processes,the complete scratches on the surface of the automotive stamped parts can be extracted.(4)Measurement of scratch size.The PCA-based defect measurement method is improved,and a scratch measurement algorithm named P-NRPCA is proposed to obtain the whole scratch’s length,width,and depth parameters.The effectiveness and stability of the scratch detection method proposed in this paper are verified through experiments,and the comparison results with other methods prove the advantages of the method in this study.